基于矩和小波变换的目标图象识别

IDENTIFING OF IMAGES OF OBJECTS USING WAVELET AND MOMENT

在图象目标的识别中 ,目标图象的不变矩对于具有旋转、尺度变换特性的目标具有良好特征匹配性能 ,但其识别准确度直接受到不变矩计算区域内的目标个数的影响 .本文首先采用一维小波变换降噪并获取图象边缘 ,结合轮廓跟踪法 (“爬虫法”)与投影法对图象进行区域分割获取各目标存在区域 ,再利用目标图象的不变矩进行目标识别 ,最后通过利用目标的主轴特性及质心变化计算目标的旋转角度及尺度变换倍率 .通过仿真实验 。

The variance with respect to translation,scaling,and rotation of invariant moments of image objects is good.The number of objects in the invariant moments computer region affects the identify veracity.To resolve the above problem,this paper presents a novel image objects identify algorithm.The algorithm first transform the gray imge with 1 D wavelet to get the edge images and reduce the noise at the same time,then combine edge searching and projection to segment and gain the region which perhaps exist objects.Consequently,identify the objects through its invariant moments and computing rotary angle,scale transform times using principal axis feature,centroid respectively in the final.Experimental results show that this algorithm can well identify the objects effectively.